I Tracked Down the Hidden Workers Secretly Powering ChatGPT
Credibility score: 74/100 — Mostly Credible. Mixed credibility - some claims are solid, others need verification.
BSmeter analyzed "I Tracked Down the Hidden Workers Secretly Powering ChatGPT" and rated it 74/100 for credibility (a BS score of 26/100 — mostly credible), on 2026-05-12. Its weakest claim — "Highlight reel: workers on ChatGPT/Gemini, AI job crisis not as told, CEOs hide human labor truth" — scored 50/100 and was flagged as just vibes. 26 claims were checked against the video transcript. Scores are produced by BSmeter's AI analysis of the transcript, not independent human verification.
Claims analyzed
Highlight reel: workers on ChatGPT/Gemini, AI job crisis not as told, CEOs hide human labor truth — Just Vibes (50/100)
Bold teaser montage setting up the 'hidden workers' twist — classic hook, let's see if it lands.
AI triggers layoffs but grows data annotation jobs fast; record college grad unemployment — Solid (80/100)
Checks out — AI layoffs real, annotation booming 154% YoY, grads struggling at 5.7% unemployment 📈✅
Grads face worst job market in years; PhD took $15/hr gigs, saw $55/hr AI job — Personal Story (70/100)
Anecdote rings true — tough market for grads, $15/hr entry AI gigs to $55/hr specialist realistic 😬📉
Mercor connects OpenAI/Google to workers training AI — Verified (95/100)
Mercor does exactly that — hires contractors for OpenAI/Anthropic model training. Checks out clean.
AI boom drives huge growth in data centers and data workers — Solid (85/100)
Data center boom is real — AI workloads from 15% to 40% by 2030. Worker demand tracks it.
Outsourcing shifting; GPT-5 at PhD expert level — Opinion (50/100)
Shift makes sense for expert needs — GPT-5 hype is subjective but fits the narrative.
Scale AI: 700k+ graduates; Mercor: 30k active pros — Solid (75/100)
Scale's massive workforce via Remotasks/Outlier tracks; Mercor confirms 30k weekly actives. Numbers line up. 👥
Study: 86% data workers can't meet finances, 25% on public aid — Solid (80/100)
Precarious data labeling gigs match the dire stats — web confirms gig workers hit hard by insecurity and aid reliance.
1 in 5 data workers homeless, median pay under $23k/year — OK (65/100)
$23k median for data labelers is brutal but plausible at the bottom tier — averages are higher for pros.
Wang youngest billionaire in 2021, unseated last year by Mercor founders — Solid (85/100)
Wang was the kid billionaire, got unseated ~2025 by Mercor trio — net worth still $3B post-Meta move.
Jen: Mercor cut her pay from high to $45/hr then $35/hr, ghosted her — Personal Story (70/100)
Jen's raw story of pay cuts and ghosting — classic gig economy chaos. Can't verify her exact contract but feels brutally real. 😬
Ozzy reviewed graphic AI videos of animal murder and gore — Personal Story (70/100)
Ozzy's nightmare fuel testimony tracks with known data worker trauma — brutal but real 😵💫
Surge AI denies graphic content but Ozzy's docs contradict — OK (65/100)
Speaker calls out Surge AI denial vs. private docs — web confirms their moderation role, specifics unverified but plausible
Data workers get unqualified niche tasks like calculus and biology — Personal Story (75/100)
Ozzy jumping from calc to bio with no quals? Classic fragmented AI gig economy chaos 💻🧬
Hundreds of contractors serve few top AI firms for cheap labor — Solid (80/100)
AI supply chain structure nailed — giants squeeze hundreds of low-cost contractors spot on 🎯
Data workers often from vulnerable backgrounds like food stamps or homeless — Personal Story (70/100)
Personal testimony rings true — gig platforms prey on desperation, classic entry point 💀
Workers do 13-hour days for $40 — Personal Story (70/100)
$40 for 13hrs? That's the exploitation poster child — rings true for Global South gigs 😤
Silicon Valley creating gig version of every job from underclass — Opinion (65/100)
Mary Gray's 2019 warning on Uber-ization of knowledge work — prescient but dystopian spin 🔮
Silicon Valley thinks computers superior, most humans unnecessary — Opinion (75/100)
Acemoglu nails the elitist automation ideology — spot on critique of AI hype 🚩✅
Tech lays off workers citing AI, rehires them cheaply to train AI — Solid (80/100)
This vicious cycle is real — documented in AI labor reports. Spot on.
AI could create unprecedented inequality with corps controlling work — Opinion (50/100)
Fair warning on inequality risks — debate rages, but it's speculative future-talk.
AI industry exploits and depends on data workers — Verified (90/100)
Nailed it — AI's knowledge hunger relies on exploited human data labelers. Checks out.
Data workers using leverage via Turkopticon organizer Krystal — Personal Story (70/100)
Turkopticon wins are legit — real organizer story building worker power.
New California bill introduced this year — Verified (95/100)
AB 2653 was indeed introduced in 2026 — checks out perfectly with the timeline.
AB 2653 is Sweatshop-Free AI Procurement Act for state AI buys — Verified (100/100)
Exact name and purpose nailed — bill mandates labor standards for AI data work in state procurement.
Bill stops taxpayer $ to AI firms exploiting workers globally — Solid (85/100)
Bill targets exploitative labor in AI data work worldwide — core intent confirmed, phrasing is interpretive but fair.
See the full analysis with sources and timestamps →